app.py

import gradio as gr
import os
import skimage
import matplotlib.pyplot as plt
from PIL import Image
import numpy as np
from collections import OrderedDict
import torch
from imagebind import data
from imagebind.models import imagebind_model
from imagebind.models.imagebind_model import ModalityType
import torch.nn as nn


device = "cpu" #"cuda:0" if torch.cuda.is_available() else "cpu"
model = imagebind_model.imagebind_huge(pretrained=True)
model.eval()
model.to(device)

def image_text_zeroshot(texts):
    labels = [texts]
    inputs = {
        ModalityType.TEXT: data.load_and_transform_text(labels, device)
    }

    with torch.no_grad():
        embeddings = model(inputs)

    # scores = (
    #     torch.softmax(
    #         embeddings[ModalityType.VISION] @ embeddings[ModalityType.TEXT].T, dim=-1
    #     )
    #     .squeeze(0)
    #     .tolist()
    # )

    score_dict = "./assets/ICA-Logo.png" #{label: score for label, score in zip(labels, scores)}

    return score_dict


def main():

    iface = gr.Interface(
        fn= image_text_zeroshot(texts),
        inputs = gr.inputs.Textbox(lines=1, label="texts"),
        outputs = gr.inputs.Image(type="filepath", label="Output image"),
        description="""...""",
        title="ImageBind",
    )

    iface.launch()



# def image_classifier(inp):
#     return {'cat': 0.3, 'dog': 0.7}

# demo = gr.Interface(fn=image_classifier, inputs="image", outputs="label")
# demo.launch()